Printable View
more then one:Quote:
Originally Posted by Sumanji
790GX:
http://resources.vr-zone.com//eventp...d/IMG_0628.jpg
And My favorite 790FX:
http://resources.vr-zone.com//eventp...d/IMG_0629.jpg
P.S
Albatron has the clone of DFI's 790GX model:
http://resources.vr-zone.com//eventp...d/IMG_0583.jpg
I heard somewhere that AMD had very little Boards based on there chipsets. And that everyone (manufactures) are putting all efforts to Intel based Boards. But it looks like there might still be some options.
that would be from anand or toms CHARGED. i dont see why any board manufacturer would not release for amd. sure intel is probably where 80% of the money is but amd will probably use this with there anti trust lawsuit with intel now cause they are focusing on getting working boards for nahealm (stupid name....)
Did you read my thread about upcoming AM2+ boards at all? Even AMD is about to release like 5~6 new boards:shrug:Quote:
Originally Posted by charged3800z24
Biostar is about to release a few as well... Ive no idea what you're talking about:shrug:
I am not even sure how to pronounce the name....Any who.. I just read this over at Overclock3D is this correct?... Has anyone one heard of any Denab or any 45nm chip talk from AMD at Computex yet?
Quote:
Originally Posted by Rammsteiner
Ya I read.. that what I was getting at. It seems the web is flooded with "AMD is boring this year" bla bla.. But it doesn't apear to be true.. I am looking forward to the new boards. I am itchy to upgrade.
Oh. here is my 9600 BE screen shot...Ipowerweb finally came back online.
Unfortunately so much negative publicity has build up around AMD, or to be more precise – I has been build that people seams to take any bad, negative, or FUDicated info about AMD for granted. With that in mind it’s no wonder that this FUD about lack of interest for AMD platform with mobo manufacturers has overshadowed the fact that many new mobos for AMD are in production, or have been launched!Quote:
Originally Posted by charged3800z24
For example did you know that there are more than twenty mainboards based on month old NVIDIA chipsets… also basically there won’t be a known name in mobo business that will not have at least two models with SB750 in upcoming months…
Nedjo: I like the layout of that DFI 790fx/750 board. The only thing that would worry me a little is the onboard cooling. Pretty sure we all know the PWM section of these boards get very hot...
I'm not sure that cooling looks beefy enough to handle a decent OC... :confused:
Any other opinions? Or did they change to components that don't produce as much heat (seems unlikely)....
i have little unstable problem, its not PSU related, because i was lucky to test 3 diffrent PSUs with teh phenom
MIST/Cooltek 600W (12v 4x20A)
Corsair 620HX (12v 3x18A)
Hiper Type-R 580W Blue
i cant get it stable at 2.8ghz
temps are around ~55c
testing 2.1v on ram atm
any recommendations? :am:
Try pushing past 2.8, When i hit 2.8 I could not get it stable. But 2.9 was fine even on stock v-core...Also there is AOD which allows independent core clocking. I suggest you leave bios stock when using AOD. Unless you modify the default config file to match your settings. There is a new version 2.1 which has been working quite well for me so far.I have notice that core 2 falls short before the others. And then core 3. Core's 0 & 1 seem to be the most solid.Quote:
Originally Posted by wiak
Nedjo... am I reading that Albatron board right? It say Hyper Stransport 3.0 @ 3.0ghz....
Dang... I hadn't seen that... I was still using 2.0.17. Downloading 2.1 now... hopefully will work better.Quote:
Originally Posted by charged3800z24
Added to TechPowerUp!
http://www.techpowerup.com/downloads...drive_2.1.html
I found it a few days ago when downloading the new ATI drivers. I saw the link for AOD and thought I'd peak to see if any new version was out. It is different then the other versions...no more yellow mode..and other changes.
didn't like the new AOD.
anyways I put a temp senors form a Sythe "Kaze master" panel in my Voltage regulators, one in the northbrige, the other two went to ram & power supply.
at load the Voltage regulator's get to about 55c north bridge gets up to about 45C. my rams pretty cool only 34C. Ps 40C.
this 9550 has only giving me one BSOD clock interrupt was not received. That was when I bumped the volts up form 1.277 to 1.350 thats when it happened. Restated & running fine now.
I'm betting my ram isn't working totally great it won't go past 430 on 1:2 for 800mhz
ganged & unganged isn't changing it either.
It's 1066mhz ram stickers say 1.9-2.3V 2gbs sticks
board keeps saying Ram over clock failed. even a 2.1 volts.
Are you setting the timings correctly?..
I like the new AOD. It allows you to drop down the HT multi so you can do better HTT OC'ing.
Right at -12c on my temp display, I didnt use bios for temp reading, the td between active core and probe is about 12c on the F1 for a quad core.Quote:
Originally Posted by msimax
I went through a lot of bios options on both Nvidia and ATI, ATI sure does have a lot of tweaking options, some I am unfamiliar with. I found the GPU clocks on the ATI board, 500Mhz stock and didnt have a chance to push.
I even listened to suggestions, tried single and all filled ram combinations.
The memory I used was OCZ Flex 9600 modules, I had no problems running all slots filled on any board nor did it limit my cpu oc even though I didnt push the memory hard enough.
I read 3Ghz too. Although it says to 3Ghz, not at 3Ghz. But still, 3Ghz... Would this mean we can expect 3Ghz+NB speeds or is it more like a :banana::banana::banana::banana:ty brag rights feature.:confused:Quote:
Originally Posted by charged3800z24
You're lucky on those VRMs. Mine run > 75C when under full load (at 1.34V) despite the large heatsink and heatpipe assembly. It's crazy that the K9A2 makes horrible contact with the MOSFETs. One is fully uncovered (so I stuck a heatsink on it). 2/3 are partially uncovered. Only one column of FETs are completely covered.
charged3800z24 AOD has always alowed you to lower the multi of you HT just not the NB. i find it looks better but the 2.1 lacks the status tab now
I'll try more relaxed timings I had them a 5-5-5- 15 (24) which they're rated for.
6-6-6-18 (???)
I have it on auto now.
Particle
My VRM's are sitting by a 120mm fan but it dosen't seem to pulling enough air threw
I'm going to get some better fans I'll add two three on motherboard soon.
My Dominators needed 7 7 7 24 (or something like that) to boot at 1066MHz memclk mode with subtimings set to auto settings. You just have to find out what settings it is stable at, and work your way down to tighter timings. Long process..
I feel your pain:eek:Quote:
Originally Posted by LIKMARK
I guess the best thing to do is buy one of these OCZ sets Tony made a custom SPD for with AM2+. AFAIK that's about the only brand offering a good solution thus far.
Im running Axe RAMs, they work very nice as well:D
However, I run them DDR800 4-4-4, but they do DDR1066 5-5-5 too. I got them even DDR1180 5-5-5. I can do even higher however it needs Drive Strength tweaking, gave up on that after 2 weeks. But some guy at TPU hit DDR1400 with only 1.35V and 5-5-5 timings.
Hehe, my point was that it probably is immature bioses that is the cause of bad memory performance/compatability in addition to picky imc on K10. You need to know what you're doing to make memory work with 1066MHz memclk (wich is why there is no suprise that Tony makes his modules run at nice speeds with low voltage). Having a board that supports memory tweaking (read: top end 790FX/nforce780) is also nice. Having budget boards with stripped bioses makes memory tweaking harder.
FYI I made my Dominators work at 5 7 7 18 2t with Abit AX78, a board not known for memory performance. I have not tried them on other boards (with Phenom, sold it..) at memclk 1066, but they run fine with Athlon X2 on Crosshair @ 5 5 5 15 2t 1000MHz 2,0v (cpu won't oc further)
Quote:
Originally Posted by LIKMARK
immature lol like you can only change one thing at time in them that doesn't bother me much.
I have msi K9a2 with bios 1.2.
I think I'll snatch on these up, for bios flash if I get annoyed enough
http://www.newegg.com/Product/Produc...82E16819103196
http://valid.x86-secret.com/show_oc.php?id=373702
:D:D:D
who is jealous ?
Congratulations! :eek2: I take it B3 is not coldbugged then?Quote:
Originally Posted by Fatalbert
Nice :shocked:
it's been do with wc and 1.424, it's not my score remember that I only deliver good info ;)
Holy :banana::banana::banana::banana:, that's awesome! And so less Vcore too.Quote:
Originally Posted by Fatalbert
Ive been slapping myself past 3 weeks NOT to buy a new Phenom just yet and wait for the SB750 boards... But I tell you, it's very hard:mad:
well rammsteiner you could always do both??? no???? you know you want 3 9850BE's and 3 boards. dont we all? no? thats a shame
Phenoms are just a boatload of fun. I don't know why. They're fun toys to play with even if you don't need a system.
Well, basicly I want to test this Phenom on a non-flaky board:D Then Ill see if I get a 2nd Phenom or not.Quote:
Originally Posted by Calcifer
http://valid.x86-secret.com/show_oc.php?id=373855
:D:D:D:D
another score.
:toast: Awesome new world record :toast:
Let's see some benches!!!
Milans told me this is not the end of a cpu ;)
sure. I hope it can overclock to 3.9-4.0 GHz on LN2 or SS but currently I don't have both ;)
Im afraid these OC's were not stable though?:(Quote:
Originally Posted by Fatalbert
maybe for valid it will be stable but for benchmarks, maybe for spi 1m :D
btw
czesc Milans ;)
no it wasn't. It is only validation. I don't know how fast it is able to work 100% stable because my WC isn't efficient enough. After 1h of IDLE temperature was 55-60*C - very, very hot.
That's indeed very hot:shocked:Quote:
Originally Posted by Milans
Hmm i see something... why is htt low like 212?is it harder to go high with higher htt?Quote:
Originally Posted by Milans
The deffrent on the scores are multi 17,5 instead of 17 = higher total clocks.
because phenoms don't like high htt (300>)
Lets runs some benchmarks together to see how ti compares to a 3.6ghz q6600.Quote:
Originally Posted by Fatalbert
I've always thought the phenom arch scaled better with mhz.
if you want some benchmarks, read what you want, maybe Milans do it
Hmmm, looks like a nice idea:DQuote:
Originally Posted by Unoid
nice idea ;) write down your propositions and give some screens ;) (you can give 3DMarks too. I don't have 8800GT right now, but it is no problem to get it)
Nice to see some higher clocks.. I am in the works making a liquid cooling deivce. I am always hitting 55+*c when running 3.1 3.1 2.9 2.9 on heavy loads.
It would also be fun to see what your individual cores are capable of (leave the other three at ~2.5GHz or reasonably close if you want to change HTT).
Maybe you can start new thread 'cos this one has gone way to long and deep and only handful of us are still in it this far!?Quote:
Originally Posted by Milans
I've said this alot but I guess I'll say it again. Cold=worse overclock. It knocked my 9600 from 3020mhz down to 2860mhz stable. It kept my first 9850 at a maximum of 2.75ghz until I put it on air and then I hit 3ghz+. I'm warning you, don't do cold. You'd need to get a water cooler like Tony's for the best setup. Only thing that cold does is increase the head room of the nb which unless you have your cpu running fast then it doesn't matter.
Nedjo - I agree but I don't have time - we must wait for a weekend
Oldguy932 - you don't have right. Cold = lower HTT but it is not crucial if you have unlocked multiplier
very nice Milans, try please tu wprime, its better to scaling benchmark for x-cores CPUs than superpi.
Well, I get board easy and thought I'd McGyver something... I am using AC freezer 64 pro's on both Phenoms, and I hit temp issues when cranking out stress test and the cores start to drop like flies.. I added a modded fan to one and pulled temps lower and stress test went longer.. All I am wondering is what is to cold..?Quote:
Originally Posted by Oldguy932
Just as a WAG (and I'll let you all know by Sunday I hope!) I'd say 40°C is safe enough. My old Opty's run 3.0 GHz at that temp and the 5000BE runs about that, too, at 3.2 GHz. Any warmer and the OC starts dropping. My bet is they could probably go down near ambient and still be OK but i have no data for that - strictly a WC'er here - I haven't got the Phenom hooked up with WC yet. :-/
Nothing like yours but I did manage a validation for 3.2 GHz (on all four cores ;)) - just a simple 200x16 ...Quote:
Originally Posted by Milans
http://valid.x86-secret.com/show_oc.php?id=352629
had to go to 1.426cpu volts in bios (1.448 in cpu-z) for 220 x 11 to get stable
northbrige is 1.3 volts 1980mhz
dropped timings back 6-6-6- 18, hyper transport to x6. Set memory to 1:1 with 2.1volts
I went to ungainged to see if changed anything no much of one.
don't know if I can get more out this 9550 might have to spring for a 9850BE
When we say cold, we mean sub zero. Its a pain in the ass to get Phenom stable or even booting at sub-zero...no matter how low the htt is.Quote:
Originally Posted by charged3800z24
I could still get up to 275htt on the 9850 under cold. Using the multi didn't change anything. I had my vapo on for 4 months testing this and you don't listen. If they did clock better under cold I wouldn't have take the vapo off to put on a TRUE. That 9850 could not under any circumstances hit 3ghz until I put in on air cooling. I had a crosshair running in another system, I put the 9850 in and I could get 3ghz to get into windows with 1.35v. Under cold, it would not boot into windows with 1.4v+. More voltage just made it less likely to get into windows. Using the multiplier didn't help. Same thing with my 9600 BE. It was stable at 2.8ghz on air with 1.3v, but under cold it wouldn't be stable higher than about 2.6.
Hey Bud.. Looking at your sig, if your still using the 1.2 Bios on the K9A2 you should seriously consider flashing to a different version. From my experience 1.2 was one of the weakest bios for this board.Quote:
Originally Posted by demonkevy666
The 1.4 release would probably serve you much better.. ;)
I've don't have a 9550, but the 3 I have don't show much improvement over 1.4v's so, it seems like you might be going higher than you need to on Vcore..
1.54 Beta supports 1066 Ram (with most modules).
1.13 Beta still offers the most flexability, but P-States are tricky and that's where you'll find the benefit for this version.
You may well have reached the limit of a 9550, but it would be worth a shot to try a better bios....
So lower htt and higher multi will make cold work? , what is max multi on the phenom 9850?Quote:
Originally Posted by Milans
Then the cold bug sounds like on the K8.
imo coldbug in K10 is the same as in K8. Lower temperature mean lower HTT, but CPU can work with higher MHz. If you have unlocked multiplier you can reduce temperature and set higher multiplier to make up waste of max HTT MHz.
~60*C = 3600 MHz
~40*C = 3700 MHz
Phenom 9850 BE = unlocked multiplier up to 25x
Hmm do low htt works fine?(i know the performence will get lower), but will htt like 160 works fine?Quote:
Originally Posted by Milans
160x25 = 4000mhz :) maybe -40c temp can make it.
I hadn't try it but I think it would work fine. Problem is that you can't do it in BIOS. Good idea but very difficult to make in practice.
Lower performance? I don't think so, but with HTT 140 MHz mem on 1066div will work at 140 * 8/3 = 373 MHZ (746)
Phenoms don't overclock the same way as K8's. You need to do a bit of reading in some of our old testing threads. I'll be waiting with an I told you so later on.
Oh and I forgot to add earlier, the chips clocked even less when I turned the heater patch from underneath the cpu off. I'm telling you that chilled water is the way to go.
water between 0C and 10C is where you get the best clocks IMO, i was running 8C and could get 3600 into XP and 3500 into vista
Hmmm, where's that ad I saw for the office water-cooler ...? ;)
I was guessing thatQuote:
Originally Posted by Daveburt714
will I loose anything on the bios ?
I'll do it with live updates form MSI then
Yeah, that's why I asked in the phase change section if there was any easy way to use my vapo to keep temps up that high, only solution was to re-gas it with a different refridgerant. I'd much rather buy a water chiller from someone than make it so I can't eventually sell the vapo.
swiftech TEC block should put your cpu in that temp range or is the idea to get the water to that temp? hmm
The idea is to be able to keep the water at that temperature without much variation so that the cpu can't go very high at load but stays above freezing at idle.
Milan:Quote:
Originally Posted by Milans
so how cold is your water set up. if its the one from the picture i have seen. It looks like an ambient set up ???? if so what is your water and ambient room temp ?
I screwed up something on my board. Put one of the sticks of ram back in to see how high I could clock with a single stick. I got a little greedy and tried going for the #1 9600 spot for cpu-z on hwbot. To do it, I needed to be able to validate at 11.5x285, and while I could get it to boot, it wouldn't load vista due to heat issues. Now a while back something happened that I could only put 1.53v max through a phenom, which I only did a few times when I dropped it to one core. Now I can get up to 1.55v again, but if I set 1.45v I only get 1.4v. Not sure why it would do that, but it almost gives me a real reason to upgrade to a SB750 rather than just because I want one.
ambient temp ~28*C. Currently CPU work at 35*C and 42*C after overclock. The reason of such high temperatures as I said before (55-60) was air in WC. I will try to reduce water temprature with "wet" ice ;)Quote:
Originally Posted by SkullCracka
new beta of overdrive out
http://www.techpowerup.com/downloads/1101/file.html
sounds fancy :PQuote:
new feature of “Advance Clock Calibration” by adding a new page named “CMOS”.
updated msi K9a2 bios to 1.4
memory auto set my ram to 5-3-3-8 (15) T1.
which is a lot tighter then older bios would do.
booted a 1066mhz but never got to windows.
then is certainly a little girl when it comes to temps. i was using the stock water block that come with my Tt bigwater cause i never got round to buying a new one and i had a limit of 2650 on core 3 before it BSOD like hell. got me a D-Tek fuzion V2 it now runs 13C cooler just by changing the cpu block and core 3 now stops at 2780 wi\hich is due to the temps which load at 48C and idel at 32C.
still not liking me :(Quote:
Originally Posted by charged3800z24
but am trying 2812mhz @ 1.408v, 48c :rolleyes:
Quote:
Originally Posted by wiak
Hmm, did you try just hitting 2.9 all Mutli and maybe go 1.43v. That is where i started and then i tried bring down the V-core to see if it needed it.. Temps were getting high.
I thought I'd do a CPU-Z verified thiong.. so here it is CPU-Z
My system taps out between 2.7 and 2.9 Ghz depending on bus and memory settings, i.e. I can run
200 x 13.5 with 1066 6/6/6/24/18 or
200 x 14.5 with 800 5/5/5/21/15 or
232 x 11.5 with Auto mem timing
Tweaking NB or HTT voltages causes quick instability of temps. Dropping HT multiplier seems to accomplish little relative to what I can achieve with HT=NB=10x.
Having (desperately) played with individual core multi's in AOD, it appears I can run Core 0 and Core 2 up to 3.0-3.1 Ghz but Core 1 taps out around 2.9 and Core 3 taps out around 2.75 Ghz (observations from quick torture tests with prime95).
System Particulars
Phenom 9850BE on k9a2 Plat v1 Bios 1.5 dual boot XP Pro or Vista 64bit
3870x2 GPU with 8.6 drivers, 8GB Corsair Dominator (haved OC'ed with less)
QUESTIONS
I'm convinced that I need to run a beta bios to drop the NB multi; the question is which one? In particular, will 1.13B, P0H, 1.3B and/or 1.3 final be able to deal with this CPU (125 Watts) and enable NB manipulation?
Alternatively, what Windows-based NB manipulation is available?
Oddball question: is "drive strength" a term that only applies to memory? I read a random post on another site that seemed to be applying the term to multi-core core manipulation? Is that jibberish?
Thanks for any feedback...
If anyone is listening, I tried 1.13B, 1.33, and P0H Bioses to no avail (only the 113 includes P-states, btw).
My 2.7-2.9 max clocks remain, regardless of voltages applied, and regardless of being able to access the NB multi...
Tomorrow I take apart the sink/mobo/cpu to look for clues (Cyberpowerpc built the thing; not me).
I suspect the answer involves a new cpu and (790GX/SB750) mobo...
You are more likely to get a response/helpful advice/suggestion in the thread for that specific motherboard.
Herbie, unless you happen to hit the chip lottery 2.7~2.9 seems to be the average for a stable 9850 overclock. My best stable OC is 2.83 using the 1.5 release bios (202x14x11x10, 1.34v on the cpu, all others on auto).Quote:
Originally Posted by Herbie21
Like any good overclocker, I do try for better on occasion ;). But when I get frustrated that setting always works, of course YMMV...
There is a quirk to the 1.5 bios (and many others) that lets you raise the NB/IMC Multi (thanks Socketman, Charged, et.all)... The quirk is that raising the HT Multi above 10x will actually leave the HT at 2k and raise the NB multi, so a setting of 11x on the HT Multi will actually result in an HTLink speed of 2k and a NB/IMC speed of 2200.
**WARNING** This works to raise the NB Multi, but I've never had any luck getting it to lower it back down again, and it can cause alot of problems if you go too high. I would suggest not raising it above 11x.
As long as your temps aren't too high theres probably nothing wrong with the build, and I'm pretty sure your right, the problem lies with the SB600. As soon as the SB750 boards become available I'll be getting one.. :up:Quote:
Originally Posted by Herbie21
Hope this helps a little...
As followup, it looks like CyberpowerPC did a crappy job installing thermal grease. The before and after photos of my cleaning efforts speak for themselves.Quote:
Originally Posted by Daveburt714
After installing some Artic Silver 5 to the interface, I've achieved huge gains in my overclocking success:
Previous max stable clock was 2.9, and I'm currently tweaking the individual CPUs in the 3.1/2.9/2.9/2.8 range with some success. Striving to hit 15,000 in 3dmark06 (best so far ~14,500).
Sorry for the fuzzy pics...
http://img.techpowerup.org/080731/LameGrease2.jpg
http://img.techpowerup.org/080731/LameGrease3.jpg
http://img.techpowerup.org/080731/Clean1.jpg
http://img.techpowerup.org/080731/Clean3.jpg
All I do is set everything back to auto in BIOS then reboot.. NB goes back to 2000mhz.